idnits 2.17.1 draft-ietf-soc-load-control-event-package-09.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 30, 2013) is 3916 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 2141 (Obsoleted by RFC 8141) ** Obsolete normative reference: RFC 3023 (Obsoleted by RFC 7303) == Outdated reference: A later version (-15) exists of draft-ietf-soc-overload-control-13 Summary: 2 errors (**), 0 flaws (~~), 2 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IETF SOC Working Group C. Shen 3 Internet-Draft H. Schulzrinne 4 Intended status: Standards Track Columbia U. 5 Expires: January 31, 2014 A. Koike 6 NTT 7 July 30, 2013 9 A Session Initiation Protocol (SIP) Load Control Event Package 10 draft-ietf-soc-load-control-event-package-09.txt 12 Abstract 14 We define a load control event package for the Session Initiation 15 Protocol (SIP). It allows SIP entities to distribute load filtering 16 policies to other SIP entities in the network. The load filtering 17 policies contain rules to throttle calls based on their source or 18 destination domain, telephone number prefix or for a specific user. 19 The mechanism helps to prevent signaling overload and complements 20 feedback-based SIP overload control efforts. 22 Status of This Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on January 31, 2014. 39 Copyright Notice 41 Copyright (c) 2013 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 57 2. Conventions . . . . . . . . . . . . . . . . . . . . . . . . . 5 58 3. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 5 59 4. Design Requirements . . . . . . . . . . . . . . . . . . . . . 6 60 5. SIP Load Filtering Overview . . . . . . . . . . . . . . . . . 6 61 5.1. Load Filtering Policy Format . . . . . . . . . . . . . . 6 62 5.2. Load Filtering Policy Computation . . . . . . . . . . . . 7 63 5.3. Load Filtering Policy Distribution . . . . . . . . . . . 7 64 5.4. Applicable Network Domains . . . . . . . . . . . . . . . 10 65 6. Load Control Event Package . . . . . . . . . . . . . . . . . 11 66 6.1. Event Package Name . . . . . . . . . . . . . . . . . . . 11 67 6.2. Event Package Parameters . . . . . . . . . . . . . . . . 11 68 6.3. SUBSCRIBE Bodies . . . . . . . . . . . . . . . . . . . . 12 69 6.4. SUBSCRIBE Duration . . . . . . . . . . . . . . . . . . . 12 70 6.5. NOTIFY Bodies . . . . . . . . . . . . . . . . . . . . . . 12 71 6.6. Notifier Processing of SUBSCRIBE Requests . . . . . . . . 12 72 6.7. Notifier Generation of NOTIFY Requests . . . . . . . . . 12 73 6.8. Subscriber Processing of NOTIFY Requests . . . . . . . . 13 74 6.9. Handling of Forked Requests . . . . . . . . . . . . . . . 14 75 6.10. Rate of Notifications . . . . . . . . . . . . . . . . . . 14 76 6.11. State Delta . . . . . . . . . . . . . . . . . . . . . . . 15 77 7. Load Control Document . . . . . . . . . . . . . . . . . . . . 15 78 7.1. Format . . . . . . . . . . . . . . . . . . . . . . . . . 15 79 7.2. Namespace . . . . . . . . . . . . . . . . . . . . . . . . 16 80 7.3. Conditions . . . . . . . . . . . . . . . . . . . . . . . 16 81 7.3.1. Call Identity . . . . . . . . . . . . . . . . . . . . 16 82 7.3.2. Method . . . . . . . . . . . . . . . . . . . . . . . 19 83 7.3.3. Target SIP Entity . . . . . . . . . . . . . . . . . . 19 84 7.3.4. Validity . . . . . . . . . . . . . . . . . . . . . . 20 85 7.4. Actions . . . . . . . . . . . . . . . . . . . . . . . . . 21 86 7.5. Complete Examples . . . . . . . . . . . . . . . . . . . . 22 87 7.5.1. Load Control Document Examples . . . . . . . . . . . 22 88 7.5.2. Message Flow Examples . . . . . . . . . . . . . . . . 25 89 8. XML Schema Definition for Load Control . . . . . . . . . . . 27 90 9. Related Work . . . . . . . . . . . . . . . . . . . . . . . . 30 91 9.1. Relationship with Load Filtering in PSTN . . . . . . . . 30 92 9.2. Relationship with Other IETF SIP Overload Control Efforts 31 93 10. Discussion of this specification meeting the requirements of 94 RFC5390 . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 95 11. Security Considerations . . . . . . . . . . . . . . . . . . . 37 96 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 37 97 12.1. Load Control Event Package Registration . . . . . . . . 37 98 12.2. application/load-control+xml MIME Registration . . . . . 38 99 12.3. Load Control Schema Registration . . . . . . . . . . . . 39 100 13. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 39 101 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 39 102 14.1. Normative References . . . . . . . . . . . . . . . . . . 39 103 14.2. Informative References . . . . . . . . . . . . . . . . . 40 104 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 41 106 1. Introduction 108 Proper functioning of Session Initiation Protocol (SIP) [RFC3261] 109 signaling servers is critical in SIP-based communications networks. 110 The performance of SIP servers can be severely degraded when the 111 server is overloaded with excessive number of signaling requests. 112 Both legitimate and malicious traffic can overload SIP servers, 113 despite appropriate capacity planning. 115 There are three common examples of legitimate short-term increases in 116 call volumes. Viewer-voting TV shows or ticket giveaways may 117 generate millions of calls within a few minutes. Call volume may 118 also spike during special holidays such as New Year's Day and 119 Mother's Day. Finally, callers may want to reach friends and family 120 in natural disaster areas such as those affected by hurricanes. When 121 possible, only calls traversing overloaded servers should be 122 throttled under those conditions. 124 SIP load control mechanisms are needed to prevent congestion collapse 125 in these cases [RFC5390]. There are two types of load control 126 approaches. In the first approach, feedback control, SIP servers 127 provide load limits to upstream servers, to reduce the incoming rate 128 of all SIP requests [I-D.ietf-soc-overload-control]. These upstream 129 servers then drop or delay incoming SIP requests. Feedback control 130 is reactive and affects signaling messages that have already been 131 issued by user agent clients. They work well when SIP proxy servers 132 in the core networks (core proxy servers) or destination-specific SIP 133 proxy servers in the edge networks (edge proxy servers) are 134 overloaded. By their nature, they need to distribute rate, drop or 135 window information to all upstream SIP proxy servers and normally 136 affect all calls equally, regardless of destination. For example, in 137 the ticket giveaway case, almost all calls to the hotline will fail 138 at the core proxy servers; if the edge proxy servers leading to the 139 core proxy servers are also overloaded, calls to other destinations 140 will also be rejected or dropped. 142 Here, we propose an additional, complementary load control mechanism, 143 called load filtering. Network operators create load filtering 144 policies that indicate calls to specific destinations or from 145 specific sources should be rate-limited or randomly dropped. These 146 load filtering policies are then distributed to SIP servers and 147 possibly SIP user agents that are likely to generate calls to the 148 affected destinations or from the affected sources. Load filtering 149 works best if it prevents calls as close to the originating user 150 agent clients as possible. 152 The load filtering approach is most applicable for situations where a 153 traffic surge and its source/destination distribution can be 154 predicted in advance. For instance, it is appropriate for a mass- 155 phone-voting event, Mother's Day, New Year's Day, and even a 156 hurricane. However, it is less likely to be effective for the 157 initial phase of unpredicted/unpredictable mass calling events, such 158 as earthquakes or terrorist attacks. In these latter cases, the 159 local traffic load may peak by more than an order of magnitude in 160 minutes, if not seconds. This does not allow time to either 161 effectively identify the load filtering policies needed, nor 162 distribute them to the appropriate servers soon enough to prevent 163 server congestion. Once other, more immediate, techniques (such as 164 the loss-based or rate-based load feedback control methods) have 165 prevented the initial congestion collapse, the load filtering 166 approach can be used to effectively control the continuing overload. 168 Performing SIP load filtering involves the following components of 169 load filtering policies: format definition, computation, distribution 170 and enforcement. This specification defines the load filtering 171 policy, distribution and enforcement in the SIP load control event 172 package built upon existing SIP event notification framework. 173 However, load filtering policy computation is out of scope of this 174 specification, because it depends heavily on the actual network 175 topology and other service provider policies. 177 It should be noted that although the SIP load filtering mechanism is 178 motivated by the SIP overload control problem, which is why this 179 specification refers extensively to parallel SIP overload control 180 related efforts, the applicability of SIP load filtering extends 181 beyond the overload control purpose. For example, it can also be 182 used to implement quality of service or other service level agreement 183 commitments. Therefore, we use the term "load control event 184 package", instead of a narrower term "overload control event 185 package". 187 The rest of this specification is structured as follows: we begin by 188 listing the design requirements for this work in Section 4. We then 189 give an overview of load filtering operation in Section 5. The load 190 control event package for load filtering policy distribution is 191 detailed in Section 6. The load filtering policy format is defined 192 in the two sections that follow, with Section 7 introducing the XML 193 document for load filtering policies and Section 8 listing the 194 associated schema. Section 9 relates this work to corresponding 195 mechanisms in PSTN and other IETF efforts addressing SIP overload 196 control. Section 10 evaluates whether this specification meets the 197 SIP overload control requirements set forth by RFC5390 [RFC5390]. 198 Finally, Section 11 presents security considerations and Section 12 199 provides IANA considerations. 201 2. Conventions 203 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 204 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 205 document are to be interpreted as described in [RFC2119]. 207 3. Definitions 209 This specification reuses the definitions for "Event Package", 210 "Notification", "Notifier", "Subscriber", "Subscription" as in 211 [RFC6665]. The following additional definitions are also used. 213 Load Filtering: A load control mechanism which applies specific 214 actions to selected loads (e.g., SIP requests) matching specific 215 conditions. 217 Load Filtering Policy: A set of zero or more load filtering rules, 218 also known as load filtering rule set. 220 Load Filtering Rule: Conditions and actions to be applied for load 221 filtering. 223 Load Filtering Condition: Elements that describe how to select loads 224 to apply load filtering actions. This specification defines the 225 "call identity", "method", "target SIP identity", and "validity" 226 condition elements (Section 7.3). 228 Load Filtering Action: An operation to be taken by a load filtering 229 server on loads that match the load filtering conditions. This 230 specification allows actions such as accept, reject and redirect 231 of loads (Section 7.4). 233 Load Filtering Server: A server which performs load filtering. In 234 the context of this specification, the load filtering server is 235 the subscriber, which receives load filtering policies from the 236 notifier and enforces those policies during load filtering. 238 Load Control Document: An XML document that describes the load 239 filtering policies (Section 7). It inherits and enhances the 240 common policy document defined in [RFC4745]. 242 4. Design Requirements 244 The SIP load filtering mechanism needs to satisfy the following 245 requirements: 247 o To simplify the solution, we focus on a method for controlling SIP 248 load, rather than a generic application-layer mechanism. 250 o The load filtering policy needs to be distributed efficiently to 251 possibly a large subset of all SIP elements. 253 o The solution should re-use existing SIP protocol mechanisms to 254 reduce implementation and deployment complexity. 256 o For predictable overload situations, such as holidays and mass 257 calling events, the load filtering policy should specify during 258 what time it is to be applied, so that the information can be 259 distributed ahead of time. 261 o For destination-specific overload situations, the load filtering 262 policy should be able to describe the destination domain or the 263 callee. 265 o To address accidental and intentional high-volume call generators, 266 the load filtering policy should be able to specify the caller. 268 o Caller and callee need to be specified as both SIP URIs and 'tel' 269 URIs [RFC3966] in load filtering policies. 271 o It should be possible to specify particular information in the SIP 272 headers (e.g., prefixes in telephone numbers) which allow load 273 filtering over limited regionally-focused overloads. 275 o The solution should draw upon experiences from related PSTN 276 mechanisms where applicable. 278 o The solution should be extensible to meet future needs. 280 5. SIP Load Filtering Overview 282 5.1. Load Filtering Policy Format 284 Load filtering policies are specified by sets of rules. Each rule 285 contains both load filtering conditions and actions. The load 286 filtering conditions define identities of the targets to be 287 filtered(Section 7.3.1). For example, there are two typical resource 288 limits in a possible overload situation, i.e., human destination 289 limits (N number of call takers) and node capacity limits. The load 290 filtering targets in these two cases can be the specific callee 291 numbers or the destination domain corresponding to the overload. 292 Load filtering conditions also indicate the specific message type to 293 be matched (Section 7.3.2), with which target SIP entity the 294 filtering policy is associated (Section 7.3.3) and the period of time 295 when the filtering policy should be activated and deactivated 296 (Section 7.3.4). Load filtering actions describe the desired control 297 functions such as limiting the request rate below a certain level 298 (Section 7.4). 300 5.2. Load Filtering Policy Computation 302 Computing the load filtering policies needs to take into 303 consideration information such as overload time, scope and network 304 topology, as well as service policies. It is also important to make 305 sure that there is no resource allocation loop, and that server 306 capacity is allocated in a way which both prevents overload and 307 maximizes effective throughput (aka goodput). In some cases, in 308 order to better utilize system resources, it may be preferable to 309 employ an algorithm which dynamically computes the load filtering 310 policies based on currently observed server load status, rather than 311 using a purely static filtering policy assignment. The computation 312 algorithm for load filtering policies is out of scope of this 313 specification. 315 5.3. Load Filtering Policy Distribution 317 For load filtering policy distribution, this specification defines 318 the SIP event package for load control, which is an "instantiation" 319 of the generic SIP event notification framework [RFC6665]. The SIP 320 event notification framework provides an existing method for SIP 321 entities to subscribe to and receive notifications when certain 322 events occur. Such a framework forms a scalable event distribution 323 architecture that suits our needs. This specification also defines 324 XML schema of a load control document (Section 7), which is used to 325 encode load filtering policies. 327 In order for load filtering policies to be properly distributed, each 328 capable SIP entity in the network SHOULD subscribe to the SIP load 329 control event package from all its outgoing signaling neighbors, 330 known as notifiers (Section 6.6). Subscription is initiated and 331 maintained during normal server operation. Signaling neighbors are 332 discovered by sending signaling messages. For instance, if A sends 333 signaling requests to B, B is an outgoing signaling neighbor of A. A 334 needs to subscribe to the load control event package of B in case B 335 wants to curb requests from A. On the other hand, if B also sends 336 signaling requests to A, then B also needs to subscribe to A. The 337 subscription of neighboring SIP entities needs to be persistent so 338 that it is in place independently of any specific events requiring 339 load filtering. Key to this is the fact that following initial 340 subscription, the notifier sends a notification without a body if no 341 load filtering policy is defined (Section 6.7), and that the 342 subscription needs to be refreshed periodically to make it 343 persistent, as described in Section 4.1 and Section 4.2 of [RFC6665]. 344 The notifier will send a notification to its subscribers each time a 345 new subscription or a subscription refresh is accepted (Section 6.7). 346 The notification request includes in its body the current load 347 filtering policies (Section 7.1) from the notifier. If no such load 348 filtering policy exists, the notification request is sent without a 349 body. The subscribers MAY terminate the subscription if it no longer 350 considers the notifiers as its signaling neighbor, e.g., after an 351 extended period of absence of signaling message exchange. However, 352 if after un-subscribing, the subscriber determines that signaling 353 with the notifier becomes active again, it MUST immediately subscribe 354 to that notifier again. 356 We use the example architecture shown in Figure 1 to illustrate load 357 filtering policy distribution based on the SIP load control event 358 package mechanism. This scenario consists of two networks belonging 359 to Service Provider A and Service Provider B, respectively. Each 360 provider's network is made up of two SIP core proxy servers and four 361 SIP edge proxy servers. The core proxy servers and edge proxy 362 servers of Service Provider A are denoted as CPa1 to CPa2 and EPa1 to 363 EPa4; the core proxy servers and edge proxy servers of Service 364 Provider B are denoted as CPb1 to CPb2 and EPb1 to EPb4. 366 +-----------+ +-----------+ +-----------+ +-----------+ 367 | | | | | | | | 368 | EPa1 | | EPa2 | | EPa3 | | EPa4 | 369 | | | | | | | | 370 +-----------+ +-----------+ +-----------+ +-----------+ 371 \ / \ / 372 \ / \ / 373 \ / \ / 374 +-----------+ +-----------+ 375 | | | | 376 | CPa1 |------------------| CPa2 | 377 | | | | 378 +-----------+ +-----------+ 379 | | 380 Service | | 381 Provider A | | 382 | | 383 ================================================================= 384 | | 386 Service | | 387 Provider B | | 388 | | 389 +-----------+ +-----------+ 390 | | | | 391 | CPb1 |------------------| CPb2 | 392 | | | | 393 +-----------+ +-----------+ 394 / \ / \ 395 / \ / \ 396 / \ / \ 397 +-----------+ +-----------+ +-----------+ +-----------+ 398 | | | | | | | | 399 | EPb1 | | EPb2 | | EPb3 | | EPb4 | 400 | | | | | | | | 401 +-----------+ +-----------+ +-----------+ +-----------+ 403 Figure 1: Example Network Scenario Using SIP Load Control Event 404 Package Mechanism 406 At initialization stage, the proxy servers first identify all their 407 outgoing signaling neighbors and subscribe to them. The neighbor 408 identification process can be performed by service providers through 409 direct provisioning, or by the proxy servers themselves via 410 progressive learning from the signaling messages sent and received. 411 Assuming all signaling relationships in Figure 1 are bi-directional, 412 after this initialization stage, each proxy server will be subscribed 413 to all its neighbors. That is, EPa1 subscribes to CPa1; CPa1 414 subscribes to EPa1, EPa2, CPa2 and CPb1, so on and so forth. The 415 following cases then show two examples of how load filtering policy 416 distribution in this network works. 418 Case I: EPa1 serves a TV program hotline and decides to limit the 419 total number of incoming calls to the hotline to prevent an overload. 420 To do so, EPa1 sends a notification to CPa1 with the specific hotline 421 number, time of activation and total acceptable call rate. Depending 422 on the load filtering policy computation algorithm, CPa1 may allocate 423 the received total acceptable call rate among its neighbors, namely, 424 EPa2, CPa2, and CPb1, and notify them about the resulting allocation 425 along with the hotline number and the activation time. CPa2 and CPb1 426 may perform further allocation among their own neighbors and notify 427 the corresponding proxy servers. This process continues until all 428 edge proxy servers in the network have been informed about the event 429 and have proper load filtering policy configured. 431 In the above case, the network entity where load filtering policy is 432 first introduced is the SIP server providing access to the resource 433 that creates the overload situation. In other cases, the network 434 entry point of introducing load filtering policy could also be an 435 entity that hosts this resource. For example, an operator may host 436 an application server that performs 800 number translation services. 437 The application server may itself be a SIP proxy server or a SIP 438 Back-to-Back User Agent (B2BUA). If one of the 800 numbers hosted at 439 the application server creates the overload condition, the load 440 filtering policies can be introduced from the application server and 441 then propagated to other SIP proxy servers in the network. 443 Case II: a hurricane affects the region covered by CPb2, EPb3 and 444 EPb4. All these three SIP proxy servers are overloaded. The rescue 445 team determines that outbound calls are more valuable than inbound 446 calls in this specific situation. Therefore, EPb3 and EPb4 are 447 configured with load filtering policies to accept more outbound calls 448 than inbound calls. CPb2 may be configured the same way or receive 449 dynamically computed load filtering policies from EPb3 and EPb4. 450 Depending on the load filtering policy computation algorithm, CPb2 451 may also send out notifications to its outside neighbors, namely CPb1 452 and CPa2, specifying a limit on the acceptable rate of inbound calls 453 to CPb2's responsible domain. CPb1 and CPa2 may subsequently notify 454 their neighbors about limiting the calls to CPb2's area. The same 455 process could continue until all edge proxy servers are notified and 456 have load filtering policies configured. 458 Note that this specification does not define the provisioning 459 interface between the party who determines the load filtering policy 460 and the network entry point where the policy is introduced. One of 461 the options for the provisioning interface is the Extensible Markup 462 Language (XML) Configuration Access Protocol (XCAP) [RFC4825]. 464 5.4. Applicable Network Domains 466 This specification MUST be applied inside a 'Trust Domain'. The 467 concept of a Trust Domain is similar to that defined in [RFC3324]. A 468 Trust Domain for the purpose of SIP load filtering is a set of SIP 469 entities such as SIP proxy servers that are trusted to exchange load 470 filtering policies defined in this specification. In the simplest 471 case, a Trust Domain is a network of SIP entities belonging to a 472 single service provider who deploys it and accurately knows the 473 behaviour of those SIP entities. Such simple Trust Domains may be 474 joined to form larger Trust Domains by bi-lateral agreements between 475 the service providers of the SIP entities. 477 The key requirement of a Trust Domain for the purpose of SIP load 478 filtering is that the behavior of all SIP entities within a given 479 Trust Domain is known to comply to the following set of 480 specifications. 482 o The mechanisms used to secure the communication among SIP entities 483 within the Trust Domain. 485 o The manner used to determine which SIP entities are part of the 486 Trust Domain. 488 o That SIP entities in the Trust Domain are compliant to SIP 489 [RFC3261] 491 o That SIP entities in the Trust Domain are compliant to this 492 document. 494 It is important to note that effectiveness of SIP load filtering 495 requires that all neighbors that are possible signaling sources 496 participate and enforce the designated load filtering policies. 497 Otherwise, a single non-conforming neighbor could make the whole 498 filtering efforts useless by pumping in excessive traffic to overload 499 the server. Therefore, the SIP server that distributes load 500 filtering policies needs to take counter-measures towards any non- 501 conforming neighbors. A simple method is to reject excessive 502 requests with 503 (Service Unavailable) response messages as if they 503 were obeying the rate. Considering the rejection costs, a more 504 complicated but fairer method would be to allocate at the overloaded 505 server the same amount of processing to the combination of both 506 normal processing and rejection as the overloaded server would devote 507 to processing requests for a conforming upstream SIP server. These 508 approaches work as long as the total rejection cost does not 509 overwhelm the entire server resources. In addition, SIP servers need 510 to handle message prioritization properly while performing load 511 filtering, which is described in Section 6.8. 513 6. Load Control Event Package 515 The SIP load filtering mechanism defines a load control event package 516 for SIP based on [RFC6665]. 518 6.1. Event Package Name 520 The name of this event package is "load-control". This name is 521 carried in the Event and Allow-Events header, as specified in 522 [RFC6665]. 524 6.2. Event Package Parameters 526 No package specific event header field parameters are defined for 527 this event package. 529 6.3. SUBSCRIBE Bodies 531 This document does not define the content of SUBSCRIBE bodies. 532 Future specifications could define bodies for SUBSCRIBE messages, for 533 example to request specific types of load control event 534 notifications. 536 A SUBSCRIBE request sent without a body implies the default 537 subscription behavior as specified in Section 6.7. 539 6.4. SUBSCRIBE Duration 541 The default expiration time for a subscription to load filtering 542 policy is one hour. Since the desired expiration time may vary 543 significantly for subscriptions among SIP entities with different 544 signaling relationships, the subscribers and notifiers are 545 RECOMMENDED to explicitly negotiate appropriate subscription duration 546 when knowledge about the mutual signaling relationship is available. 548 6.5. NOTIFY Bodies 550 The body of a NOTIFY request in this event package contains load 551 filtering policies. The format of the NOTIFY request body MUST be in 552 one of the formats defined in the Accept header field of the 553 SUBSCRIBE request or be the default format, as specified in 554 [RFC6665]. The default data format for the NOTIFY request body of 555 this event package is "application/load-control+xml" (defined in 556 Section 7). This means that when NOTIFY request body exists but no 557 Accept header field is specified in a SUBSCRIBE request, the NOTIFY 558 request body MUST contain "application/load-control+xml" format. 560 6.6. Notifier Processing of SUBSCRIBE Requests 562 The notifier accepts a new subscription or updates an existing 563 subscription upon receiving a valid SUBSCRIBE request. 565 If the identity of the subscriber sending the SUBSCRIBE request is 566 not allowed to receive load filtering policy, the notifier MUST 567 return a 403 "Forbidden" response. 569 If none of MIME types specified in the Accept header of the SUBSCRIBE 570 request is supported, the notifier SHOULD return 406 "Not Acceptable" 571 response. 573 6.7. Notifier Generation of NOTIFY Requests 575 A notifier MUST send a NOTIFY request with its current load filtering 576 policy to the subscriber upon successfully accepting or refreshing a 577 subscription. If no load filtering policy needs to be distributed 578 when the subscription is received, the notifier SHOULD sent a NOTIFY 579 request without body to the subscriber. The content-type header 580 field of this NOTIFY request MUST indicate the correct body format as 581 if the body were present (e.g., "application/load-control+xml"). 582 Sending this NOTIFY request without body is often the case when a 583 subscription is initiated for the first time, e.g., when a SIP entity 584 is just introduced, because there may be no planned events that 585 require load filtering at that time. A notifier SHOULD generate 586 NOTIFY requests each time the load filtering policy changes, with the 587 maximum notification rate not exceeding values defined in 588 Section 6.10. 590 6.8. Subscriber Processing of NOTIFY Requests 592 The subscriber is the load filtering server which enforces load 593 filtering policies received from the notifier. The way subscribers 594 process NOTIFY requests depends on the load filtering policies 595 conveyed in the notifications. Typically, load filtering policies 596 consist of rules specifying actions to be applied to requests 597 matching certain conditions. A subscriber receiving a notification 598 first installs these rules and then enforce corresponding actions on 599 requests matching those conditions, for example, limiting the sending 600 rate of call requests destined for a specific callee. 602 In the case when load filtering policies specify a future validity, 603 it is possible that when the validity time comes, the subscription to 604 the specific notifier that conveyed the rules has expired. In this 605 case, it is RECOMMENDED that the subscriber re-activate its 606 subscription with the corresponding notifier. Regardless of whether 607 this re-activation of subscription is successful or not, when the 608 validity time is reached, the subscriber SHOULD enforce the 609 corresponding rules. 611 Upon receipt of a NOTIFY request with a Subscription-State header 612 field containing the value "terminated", the subscription status with 613 the particular notifier will be terminated. Meanwhile, subscribers 614 MUST also terminate previously received load filtering policies from 615 that notifier. 617 The subscriber SHOULD discard unknown bodies. If the NOTIFY request 618 contains several bodies, none of them being supported, it SHOULD 619 unsubscribe. A NOTIFY request without a body indicates that no load 620 filtering policies need to be updated. 622 When the subscriber enforces load filtering policies, it needs to 623 prioritize requests and select those requests that need to be 624 rejected or redirected. This selection is largely a matter of local 625 policy. It is expected that the subscriber will follow local policy 626 as long as the result in reduction of traffic is consistent with the 627 overload algorithm in effect at that node. Accordingly, the 628 normative behavior in the next three paragraphs should be interpreted 629 with the understanding that the subscriber will aim to preserve local 630 policy to the fullest extent possible. 632 o The subscriber SHOULD honor the local policy for prioritizing SIP 633 requests such as policies based on message type, e.g., INVITEs 634 versus requests associated with existing sessions. 636 o The subscriber SHOULD honor the local policy for prioritizing SIP 637 requests based on the content of the Resource-Priority header 638 (RPH, [RFC4412]). Specific (namespace.value) RPH contents may 639 indicate high priority requests that should be preserved as much 640 as possible during overload. The RPH contents can also indicate a 641 low-priority request that is eligible to be dropped during times 642 of overload. 644 o The subscriber SHOULD honor the local policy for prioritizing SIP 645 requests relating to emergency calls as identified by the SOS URN 646 [RFC5031] indicating an emergency request. 648 A local policy can be expected to combine both the SIP request type 649 and the prioritization markings, and SHOULD be honored when overload 650 conditions prevail. 652 6.9. Handling of Forked Requests 654 Forking is not applicable when this load control event package 655 mechanism is used within a single-hop distance between neighboring 656 SIP entities. If communication scope of the load control event 657 package mechanism is among multiple hops, forking is not expected to 658 happen either because the subscription request is addressed to a 659 clearly defined SIP entity. However, in the unlikely case when 660 forking does happen, the load control event package only allows the 661 first potential dialog-establishing message to create a dialog, as 662 specified in Section 5.9 of [RFC6665]. 664 6.10. Rate of Notifications 666 Rate of notifications is likely not a concern for this local control 667 event package mechanism when it is used in a non-real-time mode for 668 relatively static load filtering policies. Nevertheless, if 669 situation does arise that a rather frequent load filtering policy 670 update is needed, it is RECOMMENDED that the notifier do not generate 671 notifications at a rate higher than once per-second in all cases, in 672 order to avoid the NOTIFY request itself overloading the system. 674 6.11. State Delta 676 It is likely that updates to specific load filtering policies are 677 made by changing only part of the policy parameters only (e.g. 678 acceptable request rate or percentage, but not matching identities). 679 This will typically be because the utilization of a resource subject 680 to overload depends upon dynamic unknowns such as holding time and 681 the relative distribution of offered loads over subscribing SIP 682 entities. The updates could originate manually or be determined 683 automatically by an algorithm that dynamically computes the load 684 filtering policies (Section 5.2). Another factor that is usually not 685 known precisely or needs to be computed automatically is the duration 686 of the event requiring load filtering. Therefore it would also be 687 common for the validity to change frequently. 689 This event package allows the use of state delta as in [RFC6665] to 690 accommodate frequent updates of partial policy parameters. For each 691 NOTIFY transaction in a subscription, a version number that increases 692 by exactly one MUST be included in the NOTIFY request body when the 693 body is present. When the subscriber receives a state delta, it 694 associates the partial updates to the particular policy by matching 695 the appropriate rule id (Section 7.5). If the subscriber receives a 696 NOTIFY request with a version number that is increased by more than 697 one, it knows that it has missed a state delta and needs to ask for a 698 full state snapshot. Therefore, the subscriber ignores that NOTIFY 699 request containing the state delta, and re-sends a SUBSCRIBE request 700 to force a NOTIFY request containing a complete state snapshot. 702 7. Load Control Document 704 7.1. Format 706 A load control document is an XML document that describes the load 707 filtering policies. It inherits and enhances the common policy 708 document defined in [RFC4745]. A common policy document contains a 709 set of rules. Each rule consists of three parts: conditions, actions 710 and transformations. The conditions part is a set of expressions 711 containing attributes such as identity, domain, and validity time 712 information. Each expression evaluates to TRUE or FALSE. Conditions 713 are matched on "equality" or "greater than" style comparison. There 714 is no regular expression matching. Conditions are evaluated on 715 receipt of an initial SIP request for a dialog or standalone 716 transaction. If a request matches all conditions in a rule set, the 717 action part and the transformation part are consulted to determine 718 the "permission" on how to handle the request. Each action or 719 transformation specifies a positive grant to the policy server to 720 perform the resulting actions. Well-defined mechanism are available 721 for combining actions and transformations obtained from more than one 722 sources. 724 7.2. Namespace 726 The namespace URI for elements defined by this specification is a 727 Uniform Resource Namespace (URN) ([RFC2141]), using the namespace 728 identifier 'ietf' defined by [RFC2648] and extended by [RFC3688]. 729 The URN is as follows: 731 urn:ietf:params:xml:ns:load-control 733 7.3. Conditions 735 [RFC4745] defines three condition elements: , and 736 . In this specification, we define new condition elements 737 and reuse the element. The element is not used. 739 7.3.1. Call Identity 741 Since the problem space of this specification is different from that 742 of [RFC4745], the [RFC4745] element is not sufficient for 743 use with load filtering. First, load filtering may be applied to 744 different identities contained in a request, including identities of 745 both the receiving entity and the sending entity. Second, the 746 importance of authentication varies when different identities of a 747 request are concerned. This specification defines new identity 748 conditions that can accommodate the granularity of specific SIP 749 identity header fields. The requirement for authentication depends 750 on which field is to be matched. 752 The identity condition for load filtering is specified by the element and its sub-element . The element 754 itself contains sub-elements representing SIP sending and receiving 755 identity header fields: , , and . All those sub-elements are of an extended form of the 757 [RFC4745] element. In addition to the sub-elements 758 including , , and in the [RFC4745] 759 element, the extended form adds two new sub-elements, namely, and , which will be explained later in this section. 762 The [RFC4745] and elements may contain an "id" 763 attribute, which is the URI of a single entity to be included or 764 excluded in the condition. When used in the , , and elements, this "id" value is the URI 766 contained in the corresponding SIP header field, i.e., From, To, 767 Request-URI, and P-Asserted-Identity. 769 When the element contains multiple sub- 770 elements, the result is combined using logical OR. When the , 771 , and elements contain 772 multiple or or sub-elements, the result is 773 also combined using logical OR. When the sub-element further 774 contains one or more sub-elements, or when the 775 sub-element further contains one or more sub-elements, 776 the result of each or sub-element is combined 777 using a logical OR, similar to that of the [RFC4745] 778 element. However, when the element contains multiple of the 779 , , and sub-elements, 780 the result is combined using logical AND. This allows the call 781 identity to be specified by multiple fields of a SIP request 782 simultaneously, e.g., both the From and the To header fields. 784 The following shows an example of the element, which 785 matches call requests whose To header field contains the SIP URI 786 "sip:alice@hotline.example.com", or the 'tel' URI 787 "tel:+1-212-555-1234". 789 790 791 792 793 794 795 796 798 Before evaluating call-identity conditions, the subscriber shall 799 convert URIs received in SIP header fields in canonical form as per 800 [RFC3261], except that the phone-context parameter shall not be 801 removed, if present. 803 The [RFC4745] and elements may take a "domain" 804 attribute. The "domain" attribute specifies a domain name to be 805 matched by the domain part of the candidate identity. Thus, it 806 allows matching a large and possibly unknown number of entities 807 within a domain. The "domain" attribute works well for SIP URIs. 809 A URI identifying a SIP user, however, can also be a 'tel' URI. We 810 therefore need a similar way to match a group of 'tel' URIs. 811 According to [RFC3966], there are two forms of 'tel' URIs for global 812 numbers and local numbers, respectively. All phone numbers must be 813 expressed in global form when possible. The global number 'tel' URIs 814 start with a "+". The rest of the numbers are expressed as local 815 numbers, which must be qualified by a "phone-context" parameter. The 816 "phone-context" parameter may be labelled as a global number or any 817 number of its leading digits, or a domain name. Both forms of the 818 'tel' URI make the resulting URI globally unique. 820 'Tel' URIs of global numbers can be grouped by prefixes consisting of 821 any number of common leading digits. For example, a prefix formed by 822 a country code or both the country and area code identifies telephone 823 numbers within a country or an area. Since the length of the country 824 and area code for different regions are different, the length of the 825 number prefix also varies. This allows further flexibility such as 826 grouping the numbers into sub-areas within the same area code. 'Tel' 827 URIs of local numbers can be grouped by the value of the "phone- 828 context" parameter. 830 The and sub-elements in the [RFC4745] 831 element do not allow additional attributes to be added directly. 832 Redefining behavior of their existing attribute creates 833 backward-compatibility issues. Therefore, this specification defines 834 the and sub-elements that extend the 835 [RFC4745] element. Both of them have a "prefix" attribute 836 for grouping 'tel' URIs, similar to the "domain" attribute for 837 grouping SIP URIs in existing and sub-elements. For 838 global numbers, the "prefix" attribute value holds any number of 839 common leading digits, for example, "+1-212" for U.S. phone numbers 840 within area code "212" or "+1-212-854" for the organization with U.S. 841 area code "212" and local prefix "854". For local numbers, the 842 "prefix" attribute value contains the "phone-context" parameter 843 value. It should be noted that visual separators (such as the "-" 844 sign) in 'tel' URIs are not used for URI comparison as per [RFC3966]. 846 The following example shows the use of the "prefix" attribute along 847 with the "domain" attribute. It matches those requests calling to 848 the number "+1-202-999-1234" but are not calling from a "+1-212" 849 prefix or a SIP From URI domain of "manhattan.example.com". 851 852 853 854 855 856 857 858 859 860 861 862 864 865 866 868 7.3.2. Method 870 The load created on a SIP server depends on the type of initial SIP 871 requests for dialogs or standalone transactions. The 872 element specifies the SIP method to which the load filtering action 873 applies. When this element is not included, the load filtering 874 actions are applicable to all applicable initial requests. These 875 requests include INVITE, MESSAGE, REGISTER, SUBSCRIBE, OPTIONS, and 876 PUBLISH. Non-initial requests, such as ACK, BYE and CANCEL MUST NOT 877 be subjected to load filtering. In addition, SUBSCRIBE requests are 878 not filtered if the event-type header field indicates the event 879 package defined in this specification. 881 The following example shows the use of the element in the 882 case the filtering actions should be applied to INVITE requests. 884 INVITE 886 7.3.3. Target SIP Entity 888 A SIP server that performs load filtering may have multiple paths to 889 route call requests matching the same set of call identity elements. 890 In those situations, the SIP load filtering server may desire to take 891 advantage of alternative paths and only apply load filtering actions 892 to matching requests for the next hop SIP entity that originated the 893 corresponding load filtering policy. To achieve that, the SIP load 894 filtering server needs to associate every load filtering policy with 895 its originating SIP entity. The element is 896 defined for that purpose and it contains the URI of the entity that 897 initiated the load filtering policy, which is generally the 898 corresponding notifier. A notifier MAY include this element as part 899 of the condition of its filtering policy being sent to the 900 subscriber, as below. 902 sip:biloxi.example.com 904 When a SIP load filtering server receives a policy with a element, it SHOULD record it and take it into 906 consideration when making load filtering decisions. If the load 907 filtering server receives a load filtering policy that does not 908 contain a element, it MAY still record the URI of 909 the load filtering policy's originator as the 910 information and consider it when making load filtering decisions. 912 The following are two examples of using the 913 element. 915 Use case I: the network has user A connected to SIP Proxy 1 (SP1), 916 user B connected to SIP Proxy 3 (SP3), SP1 and SP3 connected via 917 SIP Proxy 2 (SP2), and SP2 connected to an Application Server 918 (AS). Under normal load conditions, a call from A to B is routed 919 along the following path: A-SP1-SP2-AS-SP3-B. The AS provides a 920 non-essential service and can be bypassed in case of overload. 921 Now let's assume that AS is overloaded and sends to SP2 a load 922 filtering policy requesting that 50% of all INVITE requests be 923 dropped. SP2 can maintain AS as the for that 924 policy so that it knows the 50% drop action is only applicable to 925 call requests that must go through AS, without affecting those 926 calls directly routed through SP3 to B. 928 Use case II: An 800 translation service is installed on two 929 Application Servers, AS1 and AS2. User A is connected to SP1 and 930 calls 800-1234-4529, which is translated by AS1 and AS2 into a 931 regular E.164 number depending on, e.g., the caller's location. 932 SP1 forwards INVITE requests with Request-URI = "800 number" to 933 AS1 or AS2 based on a load balancing strategy. As calls to 934 800-1234-4529 creates a pre-overload condition in AS1, AS1 sends 935 to SP1 a load filtering policy requesting that 50% of calls 936 towards 800-1234-4529 be rejected. In this case, SP1 can maintain 937 AS1 as the for the rule, and only apply the 938 load filtering policy on incoming requests that are intended to be 939 sent to AS1. Those requests that are sent to AS2, although 940 matching the of the filter, will not be affected. 942 7.3.4. Validity 944 A filtering policy is usually associated with a validity period 945 condition. This specification reuses the element of 946 [RFC4745], which specifies a period of validity time by pairs of 947 and sub-elements. When multiple time periods are 948 defined, the validity condition is evaluated to TRUE if the current 949 time falls into any of the specified time periods. i.e., it 950 represents a logical OR operation across all validity time periods. 952 The following example shows a element specifying a valid 953 period from 12:00 to 15:00 US Eastern Standard Time on 2008-05-31. 955 956 2008-05-31T12:00:00-05:00 957 2008-05-31T15:00:00-05:00 958 960 7.4. Actions 962 The actions a load filtering server takes on loads matching the load 963 filtering conditions are defined by the element in the load 964 filtering policy, which includes any one of the three sub-elements 965 , , and . The element denotes an absolute 966 value of the maximum acceptable request rate in requests per second; 967 the element specifies the relative percentage of incoming 968 requests that should be accepted; the element describes the 969 acceptable window size supplied by the receiver, which is applicable 970 in window-based load filtering. In static load filtering policy 971 configuration scenarios, using the sub-element is RECOMMENDED 972 because it is hard to enforce the percentage rate or window-based 973 load filtering when incoming load from upstream or reactions from 974 downstream are uncertain. (See [I-D.ietf-soc-overload-control] 975 [RFC6357] for more details on rate-based, loss-based and window-based 976 load control.) 978 In addition, the element takes an optional "alt-action" 979 attribute which can be used to explicitly specify the desired action 980 in case a request cannot be processed. The "alt-action" can take one 981 of the following three values: "reject", "redirect" and "drop". 983 o The "reject" action is the default value for "alt-action". It 984 means that the load filtering server will reject the request with 985 a 503 (Service Unavailable) response message. 987 o The "redirect" action means redirecting the request to another 988 target. When it is used, an "alt-target" attribute MUST be 989 defined. The "alt-target" specifies one URI or a list of URIs 990 where the request should be redirected. The server sends out the 991 redirect URIs in a 300-class response message. 993 o The "drop" action means simply ignoring the request without doing 994 anything, which can in certain cases help save processing 995 capability during overload. For example, when SIP is running over 996 a reliable transport such as TCP, the "drop" action does not send 997 out the rejection response, neither does it close the transport 998 connection. The "drop" action is generally also good at dealing 999 with telephony DOS attacks. However, when running SIP over an 1000 unreliable transport such as UDP, using the "drop" action will 1001 create message retransmissions that further worsen the possible 1002 overload situation. Therefore, any "drop" action applied to an 1003 unreliable transport MUST be treated as if it were "reject". 1005 In the following element example, the server accepts 1006 maximum of 100 call requests per second. The remaining calls are 1007 redirected to an answering machine. 1009 1010 1012 100 1013 1014 1016 7.5. Complete Examples 1018 7.5.1. Load Control Document Examples 1020 This section presents two complete examples of load control documents 1021 valid with respect to the XML schema defined in Section 8. 1023 The first example assumes that a set of hotlines are set up at 1024 "sip:alice@hotline.example.com" and "tel:+1-212-555-1234". The 1025 hotlines are activated from 12:00 to 15:00 US Eastern Standard Time 1026 on 2008-05-31. The goal is to limit the incoming calls to the 1027 hotlines to 100 requests per second. Calls that exceed the rate 1028 limit are explicitly rejected. 1030 1031 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 INVITE 1046 1047 2008-05-31T12:00:00-05:00 1048 2008-05-31T15:00:00-05:00 1049 1050 1051 1052 1053 100 1054 1055 1057 1058 1060 The second example considers optimizing server resource usage of a 1061 three-day period during the aftermath of a hurricane. Incoming calls 1062 to the domain "sandy.example.com" or to call destinations with prefix 1063 "+1-212" will be limited to a rate of 100 requests per second, except 1064 for those calls originating from a particular rescue team domain 1065 "rescue.example.com". Outgoing calls from the hurricane domain or 1066 calls within the local domain are never limited. All calls that are 1067 throttled due to the rate limit will be forwarded to an answering 1068 machine with updated hurricane rescue information. 1070 1071 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 INVITE 1092 1093 2012-10-25T09:00:00+01:00 1094 2012-10-28T09:00:00+01:00 1095 1096 1097 1098 1100 100 1101 1102 1104 1105 1107 Sometimes it may occur that multiple rules in a ruleset define 1108 actions that match the same methods, call identity and validity. In 1109 those cases, the "first-match-wins" principle is used. For example, 1110 in the following ruleset, the first rule requires all calls from the 1111 "example.com" domain to be rejected. Even though the rule following 1112 that one specifies that calls from "sip:alice@example.com" be 1113 redirected to a specific target "sip:eve@example.com", the calls from 1114 "sip:alice@example.com" will still be rejected because they have 1115 already been matched by the earlier rule. 1117 1118 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 INVITE 1132 1133 2013-7-2T09:00:00+01:00 1134 2013-7-3T09:00:00+01:00 1135 1136 1137 1138 1139 0 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 INVITE 1153 1154 2013-7-2T09:00:00+01:00 1155 2013-7-3T09:00:00+01:00 1156 1157 1158 1159 1161 0 1162 1163 1164 1166 1168 7.5.2. Message Flow Examples 1170 This section presents an example message flow of using the load 1171 control event package mechanism defined in this specification. 1173 atlanta biloxi 1174 | F1 SUBSCRIBE | 1175 |------------------>| 1176 | F2 200 OK | 1177 |<------------------| 1178 | F3 NOTIFY | 1179 |<------------------| 1180 | F4 200 OK | 1181 |------------------>| 1183 F1 SUBSCRIBE atlanta.example.com -> biloxi.example.com 1185 SUBSCRIBE sip:biloxi.example.com SIP/2.0 1186 Via: SIP/2.0/TCP atlanta.example.com;branch=z9hG4bKy7cjbu3 1187 From: sip:atlanta.example.com;tag=162ab5 1188 To: sip:biloxi.example.com 1189 Call-ID: 2xTb9vxSit55XU7p8@atlanta.example.com 1190 CSeq: 2012 SUBSCRIBE 1191 Contact: sip:atlanta.example.com 1192 Event: load-control 1193 Max-Forwards: 70 1194 Accept: application/load-control+xml 1195 Expires: 3600 1196 Content-Length: 0 1198 F2 200 OK biloxi.example.com -> atlanta.example.com 1200 SIP/2.0 200 OK 1201 Via: SIP/2.0/TCP biloxi.example.com;branch=z9hG4bKy7cjbu3 1202 ;received=192.0.2.1 1203 To: ;tag=331dc8 1204 From: ;tag=162ab5 1205 Call-ID: 2xTb9vxSit55XU7p8@atlanta.example.com 1206 CSeq: 2012 SUBSCRIBE 1207 Expires: 3600 1208 Contact: sip:biloxi.example.com 1209 Content-Length: 0 1211 F3 NOTIFY biloxi.example.com -> atlanta.example.com 1213 NOTIFY sip:atlanta.example.com SIP/2.0 1214 Via: SIP/2.0/TCP biloxi.example.com;branch=z9hG4bKy71g2ks 1215 From: ;tag=331dc8 1216 To: ;tag=162ab5 1217 Call-ID: 2xTb9vxSit55XU7p8@atlanta.example.com 1218 Event: load-control 1219 Subscription-State: active;expires=3599 1220 Max-Forwards: 70 1221 CSeq: 1775 NOTIFY 1222 Contact: sip:biloxi.example.com 1223 Content-Type: application/load-control+xml 1224 Content-Length: ... 1226 [Load Control Document] 1228 F4 200 OK atlanta.example.com -> biloxi.example.com 1230 SIP/2.0 200 OK 1231 Via: SIP/2.0/TCP atlanta.example.com;branch=z9hG4bKy71g2ks 1232 ;received=192.0.2.2 1233 From: ;tag=331dc8 1234 To: ;tag=162ab5 1235 Call-ID: 2xTb9vxSit55XU7p8@atlanta.example.com 1236 CSeq: 1775 NOTIFY 1237 Content-Length: 0 1239 8. XML Schema Definition for Load Control 1241 This section defines the XML schema for the load control document. 1242 It extends the Common Policy schema in [RFC4745] in two ways. 1243 Firstly, it defines two mandatory attributes for the 1244 element: version and state. The version attribute allows the 1245 recipient of the notification to properly order them. Versions start 1246 at 0, and increase by one for each new document sent to a subscriber 1247 within the same subscription. Versions MUST be representable using a 1248 non-negative 32 bit integer. The state attribute indicates whether 1249 the document contains a full load filtering policy update, or whether 1250 it contains only state delta as partial update. Secondly, it defines 1251 new members of the and elements. 1253 1254 1261 1263 1265 1266 1267 1268 1269 1270 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1287 1289 1290 1292 1293 1294 1295 1296 1298 1299 1300 1302 1303 1304 1305 1306 1307 1309 1311 1313 1314 1315 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1329 1330 1332 1333 1334 1335 1336 1337 1338 1340 1341 1343 1344 1345 1347 1348 1349 1350 1351 1353 1354 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1368 1369 1371 1372 1373 1374 1375 1376 1377 1379 1380 1381 1383 1384 1386 1387 1388 1389 1391 1393 9. Related Work 1395 9.1. Relationship with Load Filtering in PSTN 1397 It is known that existing PSTN network also uses a load filtering 1398 mechanism to prevent overload and the filtering policy configuration 1399 is done manually except in specific cases when the Intelligent 1400 Network architecture is used [Q.1248.2][E.412]. This specification 1401 defines a load filtering mechanism based on the SIP event 1402 notification framework that allows automated filtering policy 1403 distribution in suitable environments. 1405 There are control messages associated with PSTN overload control 1406 which would specify an outgoing control list, call gap duration and 1407 control duration [Q.1248.2][E.412]. These items could be roughly 1408 correlated to the identity, action and time fields of the SIP load 1409 filtering policy defined in this specification. However, the load 1410 filtering policy defined in this specification is much more generic 1411 and flexible as opposed to its PSTN counterpart. 1413 Firstly, PSTN load filtering only applies to telephone numbers. The 1414 identity element of SIP load filtering policy allows both SIP URI and 1415 telephone numbers (through 'tel' URI) to be specified. These 1416 identities can be arbitrarily grouped by SIP domains or any number of 1417 leading prefix of the telephone numbers. 1419 Secondly, the PSTN load filtering action is usually limited to call 1420 gapping. The action field in SIP load filtering policy allows more 1421 flexible possibilities such as rate throttle and others. 1423 Thirdly, the duration field in PSTN load filtering specifies a value 1424 in seconds for the load filtering duration only, and the allowed 1425 values are mapped into a value set. The time field in SIP load 1426 filtering policy may specify not only a duration, but also a future 1427 activation time which could be especially useful for automating load 1428 filtering for predictable overloads. 1430 PSTN load filtering can be performed in both edge switches and 1431 transit switches; SIP load filtering can also be applied in both edge 1432 proxy servers and core proxy servers, and even in capable user 1433 agents. 1435 PSTN load filtering also has special accommodation for High 1436 Probability of Completion (HPC) calls, which would be similar to 1437 calls designated by the SIP Resource Priority Headers [RFC4412]. SIP 1438 load filtering mechanism also allows prioritizing the treatment of 1439 these calls by specifying favorable actions for them. 1441 PSTN load filtering also provides administrative option for routing 1442 failed call attempts to either a reorder tone [E.300SerSup3] 1443 indicating overload conditions, or a special recorded announcement. 1444 Similar capability can be provided in SIP load filtering mechanism by 1445 specifying appropriate "alt-action" attribute in the SIP load 1446 filtering action field. 1448 9.2. Relationship with Other IETF SIP Overload Control Efforts 1450 The load filtering policies in this specification consist of 1451 identity, action and time. The identity can range from a single 1452 specific user to an arbitrary user aggregate, domains or areas. The 1453 user can be identified by either the source or the destination. When 1454 the user is identified by the source and a favorable action is 1455 specified, the result is to some extent similar to identifying a 1456 priority user based on authorized Resource Priority Headers [RFC4412] 1457 in the requests. Specifying a source user identity with an 1458 unfavorable action would cause an effect to some extent similar to an 1459 inverse SIP resource priority mechanism. 1461 The load filtering policy defined in this specification is generic 1462 and expected to be applicable not only to the load filtering 1463 mechanism but also to the feedback overload control mechanism in 1464 [I-D.ietf-soc-overload-control]. In particular, both mechanisms 1465 could use specific or wildcard identities for load control and could 1466 share well-known load control actions. The time duration field in 1467 the load filtering policy could also be used in both mechanisms. As 1468 mentioned in Section 1, the load filtering policy distribution 1469 mechanism and the feedback overload control mechanism address 1470 complementary areas in the overload control problem space. Load 1471 filtering is more proactive and focuses on distributing filtering 1472 policies towards the source of the traffic; the hop-by-hop feedback- 1473 based approach is reactive and targets more at traffic already 1474 accepted in the network. Therefore, they could also make different 1475 use of the generic load filtering policy components. For example, 1476 the load filtering mechanism may use the time field in the filtering 1477 policy to specify not only a control duration but also a future 1478 activation time to accommodate a predicable overload such as the one 1479 caused by Mother's Day greetings or a viewer-voting program; the 1480 feedback-based control might not need to use the time field or might 1481 use the time field to specify an immediate load control duration. 1483 10. Discussion of this specification meeting the requirements of 1484 RFC5390 1486 This section evaluates whether the load control event package 1487 mechanism defined in this specification satisfies various SIP 1488 overload control requirements set forth by RFC5390 [RFC5390]. Not 1489 all RFC5390 requirements are found applicable due to the scope of 1490 this specification. Therefore, we categorize the assessment results 1491 into Yes (meet the requirement), P/A (Partially Applicable), No (must 1492 be used in conjunction with another mechanism to meet the 1493 requirement), and N/A (Not Applicable). 1495 REQ 1: The overload mechanism shall strive to maintain the overall 1496 useful throughput (taking into consideration the quality-of- 1497 service needs of the using applications) of a SIP server at 1498 reasonable levels, even when the incoming load on the network is 1499 far in excess of its capacity. The overall throughput under load 1500 is the ultimate measure of the value of an overload control 1501 mechanism. 1503 P/A. The goal of the load filtering is to prevent overload or 1504 maintain overall goodput during the time of overload, but it is 1505 dependent on the advance predictions of the load. If the predictions 1506 are incorrect, in either direction, the effectiveness of the 1507 mechanism will be affected. 1509 REQ 2: When a single network element fails, goes into overload, or 1510 suffers from reduced processing capacity, the mechanism should 1511 strive to limit the impact of this on other elements in the 1512 network. This helps to prevent a small-scale failure from 1513 becoming a widespread outage. 1515 N/A if load filtering policies are installed in advance and do not 1516 change during the potential overload period. P/A if load filtering 1517 policies are dynamically adjusted. The algorithm to dynamically 1518 compute load filtering policies is outside the scope of this 1519 specification, while the distribution of the updated filtering 1520 policies uses the event package mechanism of this specification. 1522 REQ 3: The mechanism should seek to minimize the amount of 1523 configuration required in order to work. For example, it is 1524 better to avoid needing to configure a server with its SIP message 1525 throughput, as these kinds of quantities are hard to determine. 1527 No. This mechanism is entirely dependent on advance configuration, 1528 based on advance knowledge. In order to satisfy Req 3, it should be 1529 used in conjunction with other mechanisms which are not based on 1530 advance configuration. 1532 REQ 4: The mechanism must be capable of dealing with elements that 1533 do not support it, so that a network can consist of a mix of 1534 elements that do and don't support it. In other words, the 1535 mechanism should not work only in environments where all elements 1536 support it. It is reasonable to assume that it works better in 1537 such environments, of course. Ideally, there should be 1538 incremental improvements in overall network throughput as 1539 increasing numbers of elements in the network support the 1540 mechanism. 1542 No. This mechanism is entirely dependent on the participation of all 1543 possible neighbors. In order to satisfy Req 4, it should be used in 1544 conjunction with other mechanisms, some of which are described in 1545 Section 5.4. 1547 REQ 5: The mechanism should not assume that it will only be 1548 deployed in environments with completely trusted elements. It 1549 should seek to operate as effectively as possible in environments 1550 where other elements are malicious; this includes preventing 1551 malicious elements from obtaining more than a fair share of 1552 service. 1554 No. This mechanism is entirely dependent on the non-malicious 1555 participation of all possible neighbors. In order to satisfy Req 5, 1556 it should be used in conjunction with other mechanisms, some of which 1557 are described in Section 5.4. 1559 REQ 6: When overload is signaled by means of a specific message, 1560 the message must clearly indicate that it is being sent because of 1561 overload, as opposed to other, non overload-based failure 1562 conditions. This requirement is meant to avoid some of the 1563 problems that have arisen from the reuse of the 503 response code 1564 for multiple purposes. Of course, overload is also signaled by 1565 lack of response to requests. This requirement applies only to 1566 explicit overload signals. 1568 N/A. This mechanism signals anticipated overload, not actual 1569 overload. However the signals in this mechanism are not used for any 1570 other purpose. 1572 REQ 7: The mechanism shall provide a way for an element to 1573 throttle the amount of traffic it receives from an upstream 1574 element. This throttling shall be graded so that it is not all- 1575 or-nothing as with the current 503 mechanism. This recognizes the 1576 fact that "overload" is not a binary state and that there are 1577 degrees of overload. 1579 Yes. This event package allows rate/loss/window-based overload 1580 control options as discussed in Section 7.4. 1582 REQ 8: The mechanism shall ensure that, when a request was not 1583 processed successfully due to overload (or failure) of a 1584 downstream element, the request will not be retried on another 1585 element that is also overloaded or whose status is unknown. This 1586 requirement derives from REQ 1. 1588 N/A to the load control event package mechanism itself. 1590 REQ 9: That a request has been rejected from an overloaded element 1591 shall not unduly restrict the ability of that request to be 1592 submitted to and processed by an element that is not overloaded. 1593 This requirement derives from REQ 1. 1595 Yes. For example, load filtering policy [Section 5.1] allows the 1596 inclusion of alternative forwarding destinations for rejected 1597 requests. 1599 REQ 10: The mechanism should support servers that receive requests 1600 from a large number of different upstream elements, where the set 1601 of upstream elements is not enumerable. 1603 No. Because this mechanism requires advance configuration of 1604 specifically identified neighbors, it does not support environments 1605 where the number and identity of the upstream neighbors are not known 1606 in advance. In order to satisfy Req 10, it should be used in 1607 conjunction with other mechanisms. 1609 REQ 11: The mechanism should support servers that receive requests 1610 from a finite set of upstream elements, where the set of upstream 1611 elements is enumerable. 1613 Yes. See also answer to REQ 10. 1615 REQ 12: The mechanism should work between servers in different 1616 domains. 1618 Yes. The load control event package mechanism is not limited by 1619 domain boundaries. However, it is likely more applicable in intra- 1620 domain scenarios than in inter-domain scenarios due to security and 1621 other concerns (See also Section 5.4). 1623 REQ 13: The mechanism must not dictate a specific algorithm for 1624 prioritizing the processing of work within a proxy during times of 1625 overload. It must permit a proxy to prioritize requests based on 1626 any local policy, so that certain ones (such as a call for 1627 emergency services or a call with a specific value of the 1628 Resource-Priority header field [RFC4412]) are given preferential 1629 treatment, such as not being dropped, being given additional 1630 retransmission, or being processed ahead of others. 1632 P/A. This mechanism does not specifically address the prioritizing of 1633 work during times of overload. But it does not preclude any 1634 particular local policy. 1636 REQ 14: The mechanism should provide unambiguous directions to 1637 clients on when they should retry a request and when they should 1638 not. This especially applies to TCP connection establishment and 1639 SIP registrations, in order to mitigate against avalanche restart. 1641 N/A to the load control event package mechanism itself. 1643 REQ 15: In cases where a network element fails, is so overloaded 1644 that it cannot process messages, or cannot communicate due to a 1645 network failure or network partition, it will not be able to 1646 provide explicit indications of the nature of the failure or its 1647 levels of congestion. The mechanism must properly function in 1648 these cases. 1650 P/A. Because the load filtering policies are provisioned in advance, 1651 they are not affected by the overload or failure of other network 1652 elements. But, on the other hand, they may not, in those cases, be 1653 able to protect the overloaded network elements (see Req 1). 1655 REQ 16: The mechanism should attempt to minimize the overhead of 1656 the overload control messaging. 1658 Yes. The standardized SIP event package mechanism [RFC6665] is used. 1660 REQ 17: The overload mechanism must not provide an avenue for 1661 malicious attack, including DoS and DDoS attacks. 1663 P/A. This mechanism does provide a potential avenue for malicious 1664 attacks. Therefore the security mechanisms for SIP event packages in 1665 general [RFC6665] and of section 10 of this specification should be 1666 used. 1668 REQ 18: The overload mechanism should be unambiguous about whether 1669 a load indication applies to a specific IP address, host, or URI, 1670 so that an upstream element can determine the load of the entity 1671 to which a request is to be sent. 1673 Yes. The identity of load indication is covered in the load filtering 1674 policy format definition in Section 5.1. 1676 REQ 19: The specification for the overload mechanism should give 1677 guidance on which message types might be desirable to process over 1678 others during times of overload, based on SIP-specific 1679 considerations. For example, it may be more beneficial to process 1680 a SUBSCRIBE refresh with Expires of zero than a SUBSCRIBE refresh 1681 with a non-zero expiration (since the former reduces the overall 1682 amount of load on the element), or to process re-INVITEs over new 1683 INVITEs. 1685 N/A to the load control event package mechanism itself. 1687 REQ 20: In a mixed environment of elements that do and do not 1688 implement the overload mechanism, no disproportionate benefit 1689 shall accrue to the users or operators of the elements that do not 1690 implement the mechanism. 1692 No. This mechanism is entirely dependent on the participation of all 1693 possible neighbors. In order to satisfy Req 20, it should be used in 1694 conjunction with other mechanisms, some of which are described in 1695 Section 5.4. 1697 REQ 21: The overload mechanism should ensure that the system 1698 remains stable. When the offered load drops from above the 1699 overall capacity of the network to below the overall capacity, the 1700 throughput should stabilize and become equal to the offered load. 1702 N/A to the load control event package mechanism itself. 1704 REQ 22: It must be possible to disable the reporting of load 1705 information towards upstream targets based on the identity of 1706 those targets. This allows a domain administrator who considers 1707 the load of their elements to be sensitive information, to 1708 restrict access to that information. Of course, in such cases, 1709 there is no expectation that the overload mechanism itself will 1710 help prevent overload from that upstream target. 1712 N/A to the load control event package mechanism itself. 1714 REQ 23: It must be possible for the overload mechanism to work in 1715 cases where there is a load balancer in front of a farm of 1716 proxies. 1718 Yes. The load control event package mechanism does not preclude its 1719 use in a scenario with server farms. 1721 11. Security Considerations 1723 Two aspects of security considerations arise from this specification. 1724 One is the SIP event notification framework-based load filtering 1725 policy distribution mechanism, the other is the load filtering policy 1726 enforcement mechanism. 1728 Security considerations for SIP event package mechanisms are covered 1729 in Section 6 of [RFC6665]. A particularly relevant security concern 1730 for this event package is that if the notifiers can be spoofed, 1731 attackers can send fake notifications asking subscribers to throttle 1732 all traffic, leading to Denial-of-Service attacks. Therefore, this 1733 SIP load filtering mechanism MUST be used in a Trust Domain 1734 (Section 5.4). But if a legitimate notifier in the Trust Domain is 1735 itself compromised, additional mechanisms will be needed to detect 1736 the attack. 1738 Security considerations for load filtering policy enforcement depends 1739 very much on the contents of the policy. This specification defines 1740 possible match of the following SIP header fields in a load filtering 1741 policy: , , and . The 1742 exact requirement to authenticate and authorize these fields is up to 1743 the service provider. In general, if the identity field represents 1744 the source of the request, it SHOULD be authenticated and authorized; 1745 if the identity field represents the destination of the request, the 1746 authentication and authorization is optional. 1748 12. IANA Considerations 1750 This specification registers a SIP event package, a new MIME type, a 1751 new XML namespace, and a new XML schema. 1753 12.1. Load Control Event Package Registration 1755 This section registers an event package based on the registration 1756 procedures defined in [RFC6665]. 1758 Package name: load-control 1759 Type: package 1761 Published specification: This specification 1763 Person to contact: Charles Shen, charles@cs.columbia.edu 1765 12.2. application/load-control+xml MIME Registration 1767 This section registers a new MIME type based on the procedures 1768 defined in [RFC6838] and guidelines in [RFC3023]. 1770 MIME media type name: application 1772 MIME subtype name: load-control+xml 1774 Mandatory parameters: none 1776 Optional parameters: Same as charset parameter application/xml in 1777 [RFC3023] 1779 Encoding considerations: Same as encoding considerations of 1780 application/xml in [RFC3023] 1782 Security considerations: See Section 10 of [RFC3023] and Section 11 1783 of this specification 1785 Interoperability considerations: None 1787 Published Specification: This specification 1789 Applications which use this media type: load control of SIP entities 1791 Additional information: 1793 Magic number: None 1795 File extension: .xml 1797 Macintosh file type code: 'TEXT' 1799 Personal and email address for further information: 1801 Charles Shen, charles@cs.columbia.edu 1803 Intended usage: COMMON 1804 Author/Change Controller: IETF SOC Working Group , as designated by the IESG 1807 12.3. Load Control Schema Registration 1809 URI: urn:ietf:params:xml:schema:load-control 1811 Registrant Contact: IETF SOC working group, Charles Shen 1812 (charles@cs.columbia.edu). 1814 XML: the XML schema to be registered is contained in Section 8. 1816 Its first line is 1818 1820 and its last line is 1822 1824 13. Acknowledgements 1826 The authors would like to thank Richard Barnes, Bruno Chatras, Martin 1827 Dolly, Keith Drage, Ashutosh Dutta, Janet Gunn, Vijay Gurbani, Volker 1828 Hilt, Geoff Hunt, Carolyn Johnson, Hadriel Kaplan, Paul Kyzivat, 1829 Salvatore Loreto, Timothy Moran, Eric Noel, Parthasarathi R, Adam 1830 Roach, Shida Schubert, Robert Sparks, Phil Williams and other members 1831 of the SOC and SIPPING working group for many helpful comments. In 1832 particular, Bruno Chatras proposed the and condition elements along with many other text improvements. 1834 Janet Gunn provided detailed text suggestions including Section 10. 1835 Eric Noel suggested clarification on load filtering policy 1836 distribution initialization process. Shida Schubert made many 1837 suggestions such as terminology usage. Phil Williams suggested 1838 adding support for delta updates. Ashutosh Dutta gave pointers to 1839 PSTN references. Adam Roach suggested RFC6665-related and other 1840 helpful clarifications. Richard Barnes made many suggestions such as 1841 referencing the Trust Domain concept of RFC3324 and the use of a 1842 separate element for 'tel' URI grouping. 1844 14. References 1846 14.1. Normative References 1848 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1849 Requirement Levels", BCP 14, RFC 2119, March 1997. 1851 [RFC2141] Moats, R., "URN Syntax", RFC 2141, May 1997. 1853 [RFC3023] Murata, M., St. Laurent, S., and D. Kohn, "XML Media 1854 Types", RFC 3023, January 2001. 1856 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 1857 A., Peterson, J., Sparks, R., Handley, M., and E. 1858 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 1859 June 2002. 1861 [RFC3688] Mealling, M., "The IETF XML Registry", BCP 81, RFC 3688, 1862 January 2004. 1864 [RFC3966] Schulzrinne, H., "The tel URI for Telephone Numbers", RFC 1865 3966, December 2004. 1867 [RFC4745] Schulzrinne, H., Tschofenig, H., Morris, J., Cuellar, J., 1868 Polk, J., and J. Rosenberg, "Common Policy: A Document 1869 Format for Expressing Privacy Preferences", RFC 4745, 1870 February 2007. 1872 [RFC6665] Roach, A., "SIP-Specific Event Notification", RFC 6665, 1873 July 2012. 1875 [RFC6838] Freed, N., Klensin, J., and T. Hansen, "Media Type 1876 Specifications and Registration Procedures", BCP 13, RFC 1877 6838, January 2013. 1879 14.2. Informative References 1881 [E.300SerSup3] 1882 ITU-T, ., "North American Precise Audible Tone Plan", 1883 E.300 Series Supplement 3 , November 1988. 1885 [E.412] ITU-T, ., "Network Management Controls", E.412-2003 , 1886 January 2003. 1888 [I-D.ietf-soc-overload-control] 1889 Gurbani, V., Hilt, V., and H. Schulzrinne, "Session 1890 Initiation Protocol (SIP) Overload Control", draft-ietf- 1891 soc-overload-control-13 (work in progress), May 2013. 1893 [Q.1248.2] 1894 ITU-T, ., "Interface Recommendation for Intelligent 1895 Network Capability Set4:SCF-SSF interface", Q.1248.2 , 1896 July 2001. 1898 [RFC2648] Moats, R., "A URN Namespace for IETF Documents", RFC 2648, 1899 August 1999. 1901 [RFC3324] Watson, M., "Short Term Requirements for Network Asserted 1902 Identity", RFC 3324, November 2002. 1904 [RFC4412] Schulzrinne, H. and J. Polk, "Communications Resource 1905 Priority for the Session Initiation Protocol (SIP)", RFC 1906 4412, February 2006. 1908 [RFC4825] Rosenberg, J., "The Extensible Markup Language (XML) 1909 Configuration Access Protocol (XCAP)", RFC 4825, May 2007. 1911 [RFC5031] Schulzrinne, H., "A Uniform Resource Name (URN) for 1912 Emergency and Other Well-Known Services", RFC 5031, 1913 January 2008. 1915 [RFC5390] Rosenberg, J., "Requirements for Management of Overload in 1916 the Session Initiation Protocol", RFC 5390, December 2008. 1918 [RFC6357] Hilt, V., Noel, E., Shen, C., and A. Abdelal, "Design 1919 Considerations for Session Initiation Protocol (SIP) 1920 Overload Control", RFC 6357, August 2011. 1922 Authors' Addresses 1924 Charles Shen 1925 Columbia University 1926 Department of Computer Science 1927 1214 Amsterdam Avenue, MC 0401 1928 New York, NY 10027 1929 USA 1931 Phone: +1 212 854 3109 1932 Email: charles@cs.columbia.edu 1934 Henning Schulzrinne 1935 Columbia University 1936 Department of Computer Science 1937 1214 Amsterdam Avenue, MC 0401 1938 New York, NY 10027 1939 USA 1941 Phone: +1 212 939 7004 1942 Email: schulzrinne@cs.columbia.edu 1943 Arata Koike 1944 NTT Service Integration Labs 1945 3-9-11 Midori-cho Musashino-shi 1946 Tokyo 184-0013 1947 Japan 1949 Phone: +81 422 59 6099 1950 Email: koike.arata@lab.ntt.co.jp